Skip to content

[Clang][NFC] Move OffloadArch enum to a generic location #137070

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
merged 5 commits into from
Apr 25, 2025
Merged

[Clang][NFC] Move OffloadArch enum to a generic location #137070

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
5 commits
Select commit Hold shift + click to select a range
924f542
[Clang] Move OffloadArch enum to a generic location and add initial s…
jzc Apr 14, 2025
91b7c9d
Address review comments
jzc Apr 24, 2025
d43ae13
Update naming
jzc Apr 24, 2025
9c36e30
Use llvm instead of std
jzc Apr 24, 2025
bd5e7a2
Remove SM2 macro
jzc Apr 24, 2025
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
109 changes: 2 additions & 107 deletions clang/include/clang/Basic/Cuda.h
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -9,6 +9,8 @@
#ifndef LLVM_CLANG_BASIC_CUDA_H
#define LLVM_CLANG_BASIC_CUDA_H

#include "clang/Basic/OffloadArch.h"

namespace llvm {
class StringRef;
class Twine;
Expand Down Expand Up @@ -54,98 +56,6 @@ const char *CudaVersionToString(CudaVersion V);
// Input is "Major.Minor"
CudaVersion CudaStringToVersion(const llvm::Twine &S);

enum class OffloadArch {
UNUSED,
UNKNOWN,
// TODO: Deprecate and remove GPU architectures older than sm_52.
SM_20,
SM_21,
SM_30,
// This has a name conflict with sys/mac.h on AIX, rename it as a workaround.
SM_32_,
SM_35,
SM_37,
SM_50,
SM_52,
SM_53,
SM_60,
SM_61,
SM_62,
SM_70,
SM_72,
SM_75,
SM_80,
SM_86,
SM_87,
SM_89,
SM_90,
SM_90a,
SM_100,
SM_100a,
SM_101,
SM_101a,
SM_120,
SM_120a,
GFX600,
GFX601,
GFX602,
GFX700,
GFX701,
GFX702,
GFX703,
GFX704,
GFX705,
GFX801,
GFX802,
GFX803,
GFX805,
GFX810,
GFX9_GENERIC,
GFX900,
GFX902,
GFX904,
GFX906,
GFX908,
GFX909,
GFX90a,
GFX90c,
GFX9_4_GENERIC,
GFX942,
GFX950,
GFX10_1_GENERIC,
GFX1010,
GFX1011,
GFX1012,
GFX1013,
GFX10_3_GENERIC,
GFX1030,
GFX1031,
GFX1032,
GFX1033,
GFX1034,
GFX1035,
GFX1036,
GFX11_GENERIC,
GFX1100,
GFX1101,
GFX1102,
GFX1103,
GFX1150,
GFX1151,
GFX1152,
GFX1153,
GFX12_GENERIC,
GFX1200,
GFX1201,
AMDGCNSPIRV,
Generic, // A processor model named 'generic' if the target backend defines a
// public one.
LAST,

CudaDefault = OffloadArch::SM_52,
HIPDefault = OffloadArch::GFX906,
};

enum class CUDAFunctionTarget {
Device,
Global,
Expand All @@ -154,21 +64,6 @@ enum class CUDAFunctionTarget {
InvalidTarget
};

static inline bool IsNVIDIAOffloadArch(OffloadArch A) {
return A >= OffloadArch::SM_20 && A < OffloadArch::GFX600;
}

static inline bool IsAMDOffloadArch(OffloadArch A) {
// Generic processor model is for testing only.
return A >= OffloadArch::GFX600 && A < OffloadArch::Generic;
}

const char *OffloadArchToString(OffloadArch A);
const char *OffloadArchToVirtualArchString(OffloadArch A);

// The input should have the form "sm_20".
OffloadArch StringToOffloadArch(llvm::StringRef S);

/// Get the earliest CudaVersion that supports the given OffloadArch.
CudaVersion MinVersionForOffloadArch(OffloadArch A);

Expand Down
128 changes: 128 additions & 0 deletions clang/include/clang/Basic/OffloadArch.h
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -0,0 +1,128 @@
//===--- OffloadArch.h - Definition of offloading architectures --- C++ -*-===//
//
// Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
// See https://llvm.org/LICENSE.txt for license information.
// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
//
//===----------------------------------------------------------------------===//

#ifndef LLVM_CLANG_BASIC_OFFLOADARCH_H
#define LLVM_CLANG_BASIC_OFFLOADARCH_H

namespace llvm {
class StringRef;
} // namespace llvm

namespace clang {

enum class OffloadArch {
UNUSED,
UNKNOWN,
// TODO: Deprecate and remove GPU architectures older than sm_52.
SM_20,
SM_21,
SM_30,
// This has a name conflict with sys/mac.h on AIX, rename it as a workaround.
SM_32_,
SM_35,
SM_37,
SM_50,
SM_52,
SM_53,
SM_60,
SM_61,
SM_62,
SM_70,
SM_72,
SM_75,
SM_80,
SM_86,
SM_87,
SM_89,
SM_90,
SM_90a,
SM_100,
SM_100a,
SM_101,
SM_101a,
SM_120,
SM_120a,
GFX600,
GFX601,
GFX602,
GFX700,
GFX701,
GFX702,
GFX703,
GFX704,
GFX705,
GFX801,
GFX802,
GFX803,
GFX805,
GFX810,
GFX9_GENERIC,
GFX900,
GFX902,
GFX904,
GFX906,
GFX908,
GFX909,
GFX90a,
GFX90c,
GFX9_4_GENERIC,
GFX942,
GFX950,
GFX10_1_GENERIC,
GFX1010,
GFX1011,
GFX1012,
GFX1013,
GFX10_3_GENERIC,
GFX1030,
GFX1031,
GFX1032,
GFX1033,
GFX1034,
GFX1035,
GFX1036,
GFX11_GENERIC,
GFX1100,
GFX1101,
GFX1102,
GFX1103,
GFX1150,
GFX1151,
GFX1152,
GFX1153,
GFX12_GENERIC,
GFX1200,
GFX1201,
AMDGCNSPIRV,
Generic, // A processor model named 'generic' if the target backend defines a
// public one.
LAST,

CudaDefault = OffloadArch::SM_52,
HIPDefault = OffloadArch::GFX906,
};

static inline bool IsNVIDIAOffloadArch(OffloadArch A) {
return A >= OffloadArch::SM_20 && A < OffloadArch::GFX600;
}

static inline bool IsAMDOffloadArch(OffloadArch A) {
// Generic processor model is for testing only.
return A >= OffloadArch::GFX600 && A < OffloadArch::Generic;
}

const char *OffloadArchToString(OffloadArch A);
const char *OffloadArchToVirtualArchString(OffloadArch A);

// Convert a string to an OffloadArch enum value. Returns
// OffloadArch::UNKNOWN if the string is not recognized.
OffloadArch StringToOffloadArch(llvm::StringRef S);

} // namespace clang

#endif // LLVM_CLANG_BASIC_OFFLOADARCH_H
1 change: 1 addition & 0 deletions clang/lib/Basic/CMakeLists.txt
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -76,6 +76,7 @@ add_clang_library(clangBasic
MakeSupport.cpp
Module.cpp
ObjCRuntime.cpp
OffloadArch.cpp
OpenCLOptions.cpp
OpenMPKinds.cpp
OperatorPrecedence.cpp
Expand Down
117 changes: 0 additions & 117 deletions clang/lib/Basic/Cuda.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -73,123 +73,6 @@ CudaVersion ToCudaVersion(llvm::VersionTuple Version) {
return CudaVersion::UNKNOWN;
}

namespace {
struct OffloadArchToStringMap {
OffloadArch arch;
const char *arch_name;
const char *virtual_arch_name;
};
} // namespace

#define SM2(sm, ca) {OffloadArch::SM_##sm, "sm_" #sm, ca}
#define SM(sm) SM2(sm, "compute_" #sm)
#define GFX(gpu) {OffloadArch::GFX##gpu, "gfx" #gpu, "compute_amdgcn"}
static const OffloadArchToStringMap arch_names[] = {
// clang-format off
{OffloadArch::UNUSED, "", ""},
SM2(20, "compute_20"), SM2(21, "compute_20"), // Fermi
SM(30), {OffloadArch::SM_32_, "sm_32", "compute_32"}, SM(35), SM(37), // Kepler
SM(50), SM(52), SM(53), // Maxwell
SM(60), SM(61), SM(62), // Pascal
SM(70), SM(72), // Volta
SM(75), // Turing
SM(80), SM(86), // Ampere
SM(87), // Jetson/Drive AGX Orin
SM(89), // Ada Lovelace
SM(90), // Hopper
SM(90a), // Hopper
SM(100), // Blackwell
SM(100a), // Blackwell
SM(101), // Blackwell
SM(101a), // Blackwell
SM(120), // Blackwell
SM(120a), // Blackwell
GFX(600), // gfx600
GFX(601), // gfx601
GFX(602), // gfx602
GFX(700), // gfx700
GFX(701), // gfx701
GFX(702), // gfx702
GFX(703), // gfx703
GFX(704), // gfx704
GFX(705), // gfx705
GFX(801), // gfx801
GFX(802), // gfx802
GFX(803), // gfx803
GFX(805), // gfx805
GFX(810), // gfx810
{OffloadArch::GFX9_GENERIC, "gfx9-generic", "compute_amdgcn"},
GFX(900), // gfx900
GFX(902), // gfx902
GFX(904), // gfx903
GFX(906), // gfx906
GFX(908), // gfx908
GFX(909), // gfx909
GFX(90a), // gfx90a
GFX(90c), // gfx90c
{OffloadArch::GFX9_4_GENERIC, "gfx9-4-generic", "compute_amdgcn"},
GFX(942), // gfx942
GFX(950), // gfx950
{OffloadArch::GFX10_1_GENERIC, "gfx10-1-generic", "compute_amdgcn"},
GFX(1010), // gfx1010
GFX(1011), // gfx1011
GFX(1012), // gfx1012
GFX(1013), // gfx1013
{OffloadArch::GFX10_3_GENERIC, "gfx10-3-generic", "compute_amdgcn"},
GFX(1030), // gfx1030
GFX(1031), // gfx1031
GFX(1032), // gfx1032
GFX(1033), // gfx1033
GFX(1034), // gfx1034
GFX(1035), // gfx1035
GFX(1036), // gfx1036
{OffloadArch::GFX11_GENERIC, "gfx11-generic", "compute_amdgcn"},
GFX(1100), // gfx1100
GFX(1101), // gfx1101
GFX(1102), // gfx1102
GFX(1103), // gfx1103
GFX(1150), // gfx1150
GFX(1151), // gfx1151
GFX(1152), // gfx1152
GFX(1153), // gfx1153
{OffloadArch::GFX12_GENERIC, "gfx12-generic", "compute_amdgcn"},
GFX(1200), // gfx1200
GFX(1201), // gfx1201
{OffloadArch::AMDGCNSPIRV, "amdgcnspirv", "compute_amdgcn"},
{OffloadArch::Generic, "generic", ""},
// clang-format on
};
#undef SM
#undef SM2
#undef GFX

const char *OffloadArchToString(OffloadArch A) {
auto result = std::find_if(
std::begin(arch_names), std::end(arch_names),
[A](const OffloadArchToStringMap &map) { return A == map.arch; });
if (result == std::end(arch_names))
return "unknown";
return result->arch_name;
}

const char *OffloadArchToVirtualArchString(OffloadArch A) {
auto result = std::find_if(
std::begin(arch_names), std::end(arch_names),
[A](const OffloadArchToStringMap &map) { return A == map.arch; });
if (result == std::end(arch_names))
return "unknown";
return result->virtual_arch_name;
}

OffloadArch StringToOffloadArch(llvm::StringRef S) {
auto result = std::find_if(
std::begin(arch_names), std::end(arch_names),
[S](const OffloadArchToStringMap &map) { return S == map.arch_name; });
if (result == std::end(arch_names))
return OffloadArch::UNKNOWN;
return result->arch;
}

CudaVersion MinVersionForOffloadArch(OffloadArch A) {
if (A == OffloadArch::UNKNOWN)
return CudaVersion::UNKNOWN;
Expand Down
Loading
Loading